On Friday 27 October 2006 22:40, Norman Walsh wrote:
> / doug <> was heard to say:
> | the node useage. I guess there are semantic distinctions between
> | the topics and the sections.
>
> The important semantic distinction, I think, between sections and
> topics is that sections are (conceptually) sequential. First you read
> section 1, then you read section 2, then you read section 2.1, etc.
> Authors writing sections in this way can write with the expectation
> that, when you're reading section 2.1, you've read all of section 1
> and the introduction at the top of section 2. Phrases like "as we saw
> in the preceding section" make perfect sense.
>
> Topics, on the other hand, are discrete units. They stand alone.
> There's no expectation that there's any particular topic "before" the
> one you're looking at or "after" it. Typically there are hyperlinks
> of various sorts to help you navigate the topics in the order that you
> need to read them. You can't refer to "the previous topic" in any
> sensible way.
My bad. What I meant to say was that, for me, there were semantic
distinctions between the hierarchical nodes, which I thought could
be classed as <topic>s and the discussion/description/use cases/typing
information relevant to each of these nodes, which are separate
self-contained components of the <topic>. These "sections" would
never appear separately from their parent node/<topic> so would
never need to be reused in the "subtopic" sense. But they are still quite
legitimate partitions of information about a topic.
> |> Anyway, your example doc instance would not be valid according to
> |> the proposed content model. By design.
> |
> | So close, and yet so far :-)
> If we allow topics and sections to mix together, then I think we'll
> have a confused mess that's very difficult to explain to users.
> Several people have expressed concerns along these lines even in the
> case where there's a clear distinction.
Ok then, could the description partitions be <simplesect>s
(rather than <section>) to be followed by further subtopics
of the parent topic? i.e. would you allow <simplesect> elements
to partition information within a topic that itself, may or may not
have many subtopics? Or is it "topics" all the way down?
For the record, the data structures in question are scientific ontologies
such as this:
http://ugo.crl.nitech.ac.jp/~ddb/CIF/skunkworks.cgi/ddl2/symmetry/space_group.html
(produced from invalid DocBook source :-)
Thanks
Doug
